* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
Download The Build-Up of Bimetallic Transition Metal Clusters
Survey
Document related concepts
Transcript
The Build-Up of Bimetallic Transition Metal Clusters By Paul R. Raithby Department of Chemistry, University of‘Cambridge, England The synthesis and reaction chemistry of high nuclearity transition metal carbonyl clusters is briejly reviewed, and new synthetic strategies leading to the “rational” synthesis of bimetallic clusters containing metal cores of over 1 nm i n dimension are described. The solid state structures of a number of usmiuml mercury, osmiumlgold and rutheniumlcopper bimetallic clusters are discussed with regard to the nature of their formation, und of their bonding and redox properties. Suggestions are made as to how the synthetic strategies can be adapted to prepare bimetallic clusters of industrially useful combinations of metals. Recent work showing that bimetallic nunuparticles prepared frum clusters are catalytically active when anchored inside mesoporous silica is also discussed. Transition metal carbonyl cluster chemistry has been an important and developing topic of research in organometallic chemistry for the last three decades (1). One of the main appeals of clusters is that they lie at the interface between ‘‘conventional” organometallic chemistry and the chemistry of colloids and of the bulk metal. Figure 1 illustrates the progression in particle size from a single atom through clusters, with metal core sizes of around 1 nm; nanoparticles, with sizes up to 100 nm; leading into the colloid regime; and then on to the bulk metal. Indeed, at what size (number of metal atoms) does a metal cluster stop behaving like an organometallic complex, with bonding properties that can be described in terms of discrete molecular orbitals, and take on metallic properties, where the bonding can be described in terms of band structure? There is no immediate answer to this question, but there is a clear progression towards the clusters taking on metal- Single metal atom Particle diameter * . ;.;.: .. Cluster roi - .*... Nanoparticle -lo2i Platinum Metals Rev., 1998, 42, (4), 146-157 Colloid 4 0 3 i lic properties as the nuclearity increases, although different sizes of cluster exhibit different types of metal-like properties under different conditions (2). One of the main thrusts of cluster chemistry at Cambridge has been to prepare ever larger transition metal clusters and to investigate their physical and chemical properties. A range of clusters containing more than ten metal atoms has now been prepared and crystallographically characterised ( 3 ) and examples in which the metal atoms “condense” to form structures corresponding to the hexagonal, cubic and bodycentred cubic packings found in bulk metal have been observed, as well as other clusters, such as [Pt19(C0)22]4(4), which exhibit five-fold symmetry packing. The diameters of the metal cores in the largest of these clusters, such as [Ni,,Pt,(CO)4aH,,]”~ (n = 5, 4) are of the order of 2 nm (5). Even larger clusters containing copper and selenium 4 Bulk metal )lo% Fig. 1 The progression in particle size from a single metal atom to the bulk metal 146 have been prepared, and the largest of these to have been crystallographicallycharacterised is [ C U ~ ~ ~ S ~ in ~ which ~ ( P the P ~selenium ~ ) ~ ~ atoms ] exhibit “ABA” stacking and the copper atoms occupy interstitial sites (6). There are also reports of transition metal clusters, for instance those containing Auss(7) and Rhss( 8 ) , Pt,09(9) and Pd,,, (10) units, and a series of palladium clusters containing up to 2000 metal atoms ( I I), which have not yet been crystallographically characterised, but which must have dimensions of the order of 4 nm. Several research groups have proposed that clusters can act as good building blocks in nanoscale architecture and thus will find application in the fabrication of single electron devices (12 ) . Small metal particles and other transition metal clusters have also been clearly shown to form densely packed monolayers on electron microscope grids when they are ligated by organic surfactant molecules (13). From the viewpoint of catalysis, metal clusters can be considered as fragments of a metal surface surrounded by a layer of “adsorbed” ligand molecules. Even the largest osmium cluster carbonyl so far characterised, [Os,,(CO),]’~, where the metal framework is approximately 0.9 x 0.9 x 0.9 nm in size, contains only surface atoms, with each osmium atom being bonded to at least one carbonyl ligand, see Figure 2, (14). However, can clusters be regarded as good models of metal surfaces in heterogeneous catalytic reactions? The “cluster/surface” analogy was pointed out quite early in the development of cluster chemistry (15), and ever since then clusters have been used as models for catalytic systems (1 6). Certainly, organic molecules bond to catalytically active metal surfaces in the same way as to metal clusters, and analysis of cluster systems is easier because they can be subjected to the full range of solution spectroscopic techniques, such as IR and multinuclear NMR spectroscopies, mass spectrometry and, in the solid state, single crystal X-ray crystallography, whereas analysing the bonding modes of co-ordinated molecules on a metal surface Platinum Metals Rev., 1998, 42, (4) Fig. 2 The metal core structure of showing the four triangular faces of the Osmtetrahedron [Os,(CO),]*- under catalytic conditions is rather more challenging. However, for the majority of cluster models, for example [Os,,(CO),,]’- (14), only “surface” atoms are present, and since the sub-layers of the bulk metal influence the chemistry of the surface atoms on a catalytic surface, some aspects of the model are not valid, so the “analogy” should be treated with caution. Therefore, the simple answer to the question of whether clusters are good models for heterogeneous catalysts would be “no”. “Rational” Synthesis of High Nuclearity Mixed-Metal Clusters With a view to preparing precursor materials that could have applications in catalysis and in nanoparticle technology, the “ c l ~ ~ t egroup r” Cambridge has been developing strategies for synthesising high nuclearity mixed-metal clusters containing ten or more metal atoms in their cluster cores. Even the smallest of these clusters should have a core diameter in excess of 0.5 nm and have the advantage, unlike bimetallic particles prepared by other routes, that the exact ratio of the two metallic elements is known. As much of the early cluster synthesis work 147 I Fig. 3 The formation of the “spiked”-triangular cluster [Os,H,(CO),,] from the reaction of the activated cluster [OS~(CO)~~(M~CN)~] with [OsHZ(CO)d] involved pyrolysis or thermolysis techniques and resulted in a range of products, all in low yields, fiom a single reaction, it has been necessary to develop synthetic strategies where one target cluster molecule can be obtained in good yield (17). In order to achieve this, two fairly straightforward synthetic routes are available, given that the starting materials for the production of high nuclearity clusters are usually low nuclearity carbonyl clusters. The first route, illustrated in Figure 3 by the formation of [ O S ~ H ~ ( C Oinvolves ) ~ ~ ] , the activation of the binary carbonyl [Os,(CO),,] with Me,NO, in the presence of MeCN and results in the oxidation of carbonyl ligands to carbon dioxide and the occupation of the vacant coordination sites with labile MeCN ligands. The subsequent addition of the neutral mononuclear complex [OSH,(CO)~]displaces the MeCN groups and affords the “spiked”-triangular clus(18). In this method the ter [OS,H~(CO),~] cluster nuclearity is increased by one, by the reaction of a neutral, activated, low nuclearity M = Ru, 0 5 cluster with a neutral monometal complex. The second route involves the ionic coupling reaction between a carbonyl cluster anion and a monometal cationic species, again to increase the cluster nuclearity by one. In Figure 4, the tetranuclear osmium cluster [Os,H,(CO) ,,I is initially reduced with WPh,CO to form the dianion [Os,H,(CO) ,,I ‘-, and then treated immediately with the labile cation [M(q6C6H6)(MeCN),I2+ (M = Ru, 0 s ) to form the pentanuclear, neutral cluster [Os&W,(CO),,(q6C6H6)](1 9). By choosing appropriate cation and anion charges and ratios, the cluster nuclearity can be increased by two units, as in the reaction of [Os,(CO),,]’~with two equivalents of [Ru(q5-CsH5)(MeCN),]’to give the pentanuclear cluster [Os3Ru,(CO),,(q’-C,H,),] (20). This latter ionic coupling route has been particularly successful, and has been extensively exploited to prepare a wide range of higher nuclearity clusters containing carbocyclic ligand groups (21). The method has also been used by a number of research teams for synthesisingmixed-metal a Insertion NCMe ..!. \ (Oc)2 [H,Os,CCO),] (C0)3 H ,,OS,,M(CO)~I(C~H~) (M = Ru, 0 s ) from the coupling of Fig. 4 The synthesis of [OsdMH,(CO),,(~6-C6H6)] [OS~H~(CO with ) ~ ~[M(t16-C6H6)(MeCN),]2+ ]~~ (M = Ru, 0 s ) Platinum Metals Rev., 1998,42, (4) 148 % P b Fig. 5 The structure of the “raft” cluster [{Os,(CO)rlHg}s] showing the linking ofthree Os, triangles to the central Hg3 triangle clusters containing the coinage metals by the reaction of carbonyl cluster anions with cationic copper, silver and gold complexes (22). Osmium-MercuryClusters So far, our most extensive series of studies into the formation of mixed-metal clusters containing ten or more metal atoms have also involved the reaction of a range of ruthenium and osmium cluster carbonyl anions with late transition metal cations, such as [ H a ] ’ (X = C1, CF,), [AuPR,]’ (R = alkyl, aryl) and [CU(NCM~)~]’. The first indication that cluster build-up could occur to produce higher nuclearity clusters came from the metathesis reaction of [OS,H(CO)~~]~ with mercury(I1) salts. The product of the reaction was the extremely photolabile, dodecanuclear ‘‘rafl” cluster [ { Os,(CO) IHg},] shown in Figure 5 (23). It is significant that the three mercury atoms form a central triangle with Hg-Hg distances in the range 3.08-3.12 A. The “OS,(CO)~~” fragments bridge this central triangle, with each “Os(CO),” group forming bonds to two mercury atoms (0s-Hg in the range 2.71-2.76 A); the co-ordinated “Os(CO),” groups each form one bond to a mercury atom (0s-Hg in the range 2.98-3.05 A). This cluster can be viewed as a model for a bimetallic surface, and while being only one layer thick Platinum Metals Rev., 1998,42, (4) has a core diameter in the surface plane of approximately 1.2 nm (12 A). This coupling process, more correctly called redox condensation, resulting in cluster buildup, can also be achieved between the carbidostabilised, decanuclear cluster anion [OsloC(CO)z4]z~ and [Hg(O,SCF,),] to afford ,Figure 6. Here the anion [ { OsloC(CO)z4}zHg]2 the mercury atom links the two decanuclear clusters by bridging an edge of each Oslounit, forming a cluster containing 2 1 metal atoms (24). In this reaction, the appropriate choice of the mercury(I1) salt, [Hg(O,SCF,),], is the key to the formation of the higher nuclearity mixedmetal cluster. This is partly because the decaosmium dianion precursor has low nucleophilicity (the negative charge being delocalised over the ten metal centres) and because there is an increased propensity for the reaction to follow alternative pathways involving partial degradation or rearrangement of the metal L b Fig. 6 The molecular structure of the mercury linked cluster dianion [{Os,,C(CO),,},Hg]*~ 149 0 0 0 4 P L P 4 The Hg-Hg bonds in the Hg, dianion are an average 2.927 A long, but in the HgZdianion [Os,,Hg2C,(CO)42]2~, which has also been crystallographically characterised and found to have a pair of mercury atoms linking the two Os, units, the Hg-Hg bond is significantly shorter at 2.745 A long. Osmium-Gold Clusters Fig. 7 The structure of the dianion [MisHgX~(Co)u]’- framework. Again, the metal core in this cluster is somewhat asymmetric but has a maximum dimension of approximately 1.6 nm. When a similar reaction occurs between [M,,C(CO),,]’~(M = Os, Ru) and the mercury salt [Hg(O,CCF,),], a different type of product, [M18Hg3CZ(CO)IZ]2~, is obtained (24, 25), although the metal core still contains 21 metal atoms. This also contains a central Hg, triangle, Figure 7, linking two “M9C(CO)ZI” units, derived from the framework of the tetracapped octahedral starting materials [MI&(CO)zr]2by the loss of an “Os(CO),” vertex. As in ,I it is of interest that the mer[{OS,(CO),~H~} cury atoms have linked together between two osmiumhthenium cores forming, in metallurgical terms, a mercury domain. Another fascinating feature of the cluster dianion [ O S , , H ~ , C ~ ( C O ) is ~ ~that ] ~ - it undergoes reversible photochemical and redox extrusion of mercury atoms to give a complete series of high nuclearity clusters with general formula [Os,sHg,Cz(CO),2]”~ (n = 1-3, m = 1-4) (26). Platinum Metals Rev., 1998, 42, (4) A wide range of reactions which form high nuclearity osmium-gold clusters has been carried out (27), but this discussion will be restricted to two key syntheses which yield important information about the build-up processes in high nuclearity clusters. The first reaction is that between [OS,~C(CO),,]~~ and the polygold cation [(AuPR,),O]+(PR, = PCy,, PPh,, PMeJ’h) which affords the fourteen atom cluster [OsloC(CO) ,,Au(AuPR,) ,] (28). The structure of the PCy, derivative has been characterised crystallographically and is shown in Figure 8. The tetracapped octahedral geometry of the parent Os,, dianion is retained, and the four gold atoms form a tetrahedral cluster which is linked to the osmium core via one gold atom that bridges an 0 s - 0 s edge of one of the Q I) \8P Fig. 8 The molecular structure of [OS~~C(CO),~U(A~PC~~):~] 150 Fig. 9 (left) The molecular structure of the osmium-goldcomplex [Osla(CO),,(AuPPhzMe),] complex when it is viewed from Fig. 10 (right) The structure of the [OS,~(CO)~,(AUPP~~M~),] one end of the cluster. The novel tubular nature of the metal cnre composed of osmium atoms is clearly visible tetrahedral caps. The mean Au-Au distance to indicate a bonding interaction, while the truns within the Au, tetrahedron is only 2.7 1 A, which axial osmium atoms [Os(2)...Os(3a) and suggests that the Au-Au bonds are relatively Os(3)...Os(2a)l have moved closer together than strong. As also observed for the osmium-mer- would be expected in an octahedron, to an avercury clusters, the gold atoms have a tendency age distance of 3.3 A. The pairs of gold atoms to “cluster” together to form a domain, and are separated by 4.43 A, and the length of the do not become incorporated into the osmium tube including the gold phosphine groups atom framework. exceeds 1 nm. In order to support this type of In the second key reaction the non-carbido geometry the metal bonding must be delocalised decaosmium cluster dianion [0~10(c0)26]’~ is in character. further reduced with KPh,CO, presumably to give a tetra-anion, that is treated in situ with Ruthenium-Copper Clusters Of all the cations discussed above, [AuPPh,Me]+ to give a new type of high [Cu(NCMe),]’ is the most versatile for use in nuclearity cluster [Oslo(CO),,(AuPPh2Me),], see Figure 9 (29). cluster build-up reactions. It has been used in At first sight, from the structure of this four- combination with a number of ruthenium clusteen atom cluster, it appears that the four ter anions to produce a range of novel, high [AuPPh,Me]+cations merely cap the four end nuclearity, mixed copper-ruthenium clusters. faces of a bioctahedral osmium core, but the For example, in dichloromethane, the view looking from one end of the cluster to the reaction of the octahedral ruthenium anion other, Figure 10, shows that the octahedra are [ R U ~ ( ~ ~ - H ) ( C Owith ) , ~ ] ~an excess of distorted and that a novel tubular structure has [Cu(NCMe),]’ affords the dianionic cluster formed. The 0 s - 0 s equatorial edges [Os(4)- [{Ru,H(C0)12)2Cu7C1,]2~, see Figure 11 (30). Os(5a), Os(1)-Os(la), Os(5)-0s(4a)] have The fifteen-atom cluster core contains two expanded to lie in the range 3.29-3.32 A, a dis- Ru, tetrahedra linked through a Cu7unit which tance that is significantly longer than is judged may be described as two fused square-based Platinum Metals Rev., 1998, 42, (4) 151 Fig. 11 The structure of the [{RulH(CO)lr}rCu,Cld]2~ dianion showing the central Cu: unit - v CIO) pyramids sharing a common triangular face. Three chloro ligands each symmetrically bridge pairs of copper atoms, while the remaining copper atom, Cu(5), forms no bonds with ligands but has eight metal contacts. Thus, the immediate environment around Cu(5) is similar to that in metallic copper and, overall, again it is seen that the element with the formal d'O electron configuration has formed the central domain, and the transition metal units are fused to its periphery. The product of this reaction is very sensitive to the nature of the solvent used. In the presence of [(Ph,P),N]Cl in MeCN, when [Ru,H(CO),J is treated with a large excess of [Cu(NCMe),]+, a different product, [{RU,H(CO),,),CU~C~,]~, is obtained in good yield, see Figure 12 (30). In this case degrada- O(21I Platinum Metals Rev., 1998, 42, (4) tion of the original Ru, octahedron has not occurred, and the two octahedra are linked through two Cu, tetrahedra which share a common edge, generating two butterfly arrangements which bond to the ruthenium units. Overall, the eighteen-atom cluster can be viewed as a linear condensation of four octahedra. Two edges of the Cu, unit are symmetrically bridged by chloride ions. For both reactions, the presence of chloride ions is apparently necessary, even if, in the first case, they are abstracted from the solvent. However, by simply altering the solvent, from dichloromethane to [(Ph,P),N] C1 in MeCN, good yields of high nuclearity clusters with ruthenium:copper ratios of 8:7 (approximately 1: 1) and 2:1, respectively, are obtained from a room temperature reaction. Fig. 12 The structure of the [ {Ru,H(CO)17}2C~CI,]2~ dianion 152 Fig. 13 The core geometry in the [ {R~~OHZ(CO)~~}~CU,CI~]~ dianion This synthetic methodology can be expanded nuclearity cluster, was obtained in quantitative further. The reaction of the decanuclear dian- yield, see Figure 14 (32). The two Ru, octaheion [RU,,H,(CO)~~]~~ with excess [CU(NCM~)~]+,dra are linked by a rectangular planar arrangein dichloromethane, in the presence of chloride ment of four copper atoms, opposite edges of ions, affords the twenty-six-atom cluster which are bridged by chloride ions. Thus, with [ {RuloH,(CO),,}2C~~ClZ]~~ with a 70 per cent chloride ions present in the starting material the yield and a ruthenium:copper ratio of 10:3 (31). cluster build-up is not so efficient, resulting in Here, the two Rule units are fused on either side a sixteen-atom cluster with a ruthenium:copper of a Cu, unit which adopts the same geometry ratio of 3: 1, but the yields of the product are as that found in [ { Ru,H(CO) !,} ,CU,CI,]~- improved. In all of the copper-ruthenium clusters inves(Figure 12). The overall core geometry can be described as six fused octahedra with an addi- tigated, the copper atoms condense to form a tional ruthenium atom at each end capping a central domain and the ruthenium cluster units butterfly face to form a trigonal bipyramid, condense around the periphery to produce nanoFigure 13; this metal framework is over 1.6 sized particles based on the fusion of octahedral or tetrahedral units, just as is observed in other nm long. In order to investigate the role of the high nuclearity clusters and in close packed metchloride ions in these reactions, the carbido als (3). The advantage of the strategy employed dianion [ R U ~ C ( C O ) ~ ~was ] * - treated with in the synthesis of these copper-ruthenium CuCl, instead of the [Cu(NCMe)J+ cation, clusters is that by carefully controlling the reacanother high tion conditions, nanosized particles with and [ {Ru,C(CO),,)2Cu,Cl,]2~, Fig. 14 The structure of the dianion [{ h C ( C0)w) iC~,Clz] * Platinum Metals Rev., 1998, 42, (4) 153 Fig. 15 The network of reversible redox transformations of [Os,,Hg,,C,(CO),,]”’(11 = 1-3, m = M),Fr = ferroeene the species [OS,,H~,C~(CO)~,]”’ (n = 1-3, m = 1-4) (26) have been characterised, and it is not possible to assign realistic oxidation states for the individual mercury atoms in these species. For example, in [OS~,H~,C~(CO),,]~-, in order to balance the formal -4 charge from each “OS,C(CO)~,”fragment, each mercury atom would have to be assigned an unrealistic oxidation state of +3. The network of reversible redox transformations shown in Figure 15 confirms that this series of clusters can act as “electron sinks” (26) with a description of the bonding within the metal core best described in terms of delocalised molecular orbitals. Clusters as Nanoparticle Precursors For the chemistry outlined above it is clear that it is now possible to prepare a wide variparticular ruthenium:copper ratios can be ety of bimetallic, nanosized clusters in good yield obtained in high yields. This control has not with specific, known ratios of the two metals. been evident in previous studies (3). In principle, this strategy can be applied without difficulty to other combinations of transiSome Electrochemical tion metal atoms, such as palladium-rhodium Considerations If higher nuclearity bimetallic clusters become and platinum-rhodium, and high nuclearity “metallic” in character they would be expected clusters could be prepared. However, would to undergo extensive redox chemistry, with the these systems have real uses with industrial applicluster cores behaving as “electron sinks”. For cations? It has long been established that metal clusall of the copper-ruthenium clusters described above it is possible to assign a formal oxidation ters can be anchored to oxide surfaces such as state of + 1 to each of the copper atoms in the silica and alumina, and even encapsulated in structures, despite the view that the bonding zeolite cages; after heating the residual metal must be delocalised over the metal core. particles have high catalytic activity (33). It is Similarly, for the tubular osmium-gold clus- also known that bimetallic catalysts often exhibit ter [Osin(CO)?,(AuPPh,Me),](Figure 9) each superior operating stability to monometal sysgold atom can be assigned an oxidation state tems, and experiments have shown that rheniumof + 1 (29). Cyclic voltammetry studies show platinum clusters, supported on alumina, are that this cluster undergoes two reversible one- effective catalysts for naphtha reforming (34). electron reductions, indicating that there is However in these experiments the exact nature no major change in core geometry with the of the cluster precursors and the catalytically uptake of two electrons (27). This is consistent active materials were not known. Recently, Shapley and co-workers have prewith the cluster being able to act as an “electron sink” since it is able to reversibly take up pared a set of supported bimetallic catalysts from and release a pair of electrons repeatedly, with- two structural isomers of [RejIrC(C0)2,]’ by out it degrading, or without any major struc- deposition onto high surface area alumina and tural change that would cause the redox process activation in dihydrogen at 773 K (35). The specific activities of the catalysts depend on both to become irreversible. The situation for the osmium-mercury sys- the metal framework structure and the countems is more complicated. As noted earlier, all terion present in the precursor (either [NEt,]’ Platinum Metals Rev., 1998, 42, (4) 154 or [N(PPh,)2]+).Interpretation of EXAFS data has enabled specific models to be developed for the catalyst particle nanostructures which correlate with their catalytic activities. The more active catalysts are modelled by a hemisphere of close packed metal atoms, with an average diameter of 1 nm, with iridium at the core. In a series of related studies, Shapley has also shown that [PtRu,C(CO),,] can be used as a neutral cluster precursor for the formation of carbonsupported platinum-ruthenium nanoparticles with exceptionally narrow size and composition distributions (36). The bimetallic particles are obtained by reduction of the carbido cluster with hydrogen. A detailed structural model of the nanoparticles was deduced on the basis of in situ EXAFS, scanning transmission electron microscopy, microprobe energy-dispersive Xray analysis and electron microdifiaction studies. These experiments show that the nanoparticles have a Ru:Pt ratio of 5:1, an average diameter of approximately 1.5 nm and adopt a face centred cubic close packed structure. This is in contrast to the stable phase of the bulk alloy which is hexagonal close packed. The EXAFS studies also show that there is a non-statistical distribution of different metal atoms in the nanoparticles: the platinum atoms exhibit preferential migration to the surface of the particles under an atmosphere of dihydrogen. Of particular interest is a recent report by Johnson, Thomas and colleagues that a cluster anion [ A ~ , R u , ~ C ~ ( C O ) , , C (Figure ~ ] ~ - 16) with a structure in which a central Ag, triangle links two Ru, square based pyramids, closely related to the copper-ruthenium systems described above, had been successfullyanchored inside mesoporous silica (37). Activation and anchoring of the adsorbed cluster on the MCM41 silica support was achieved by heating the sample under dynamic vacuum. EXAFS spectroscopy confirmed the presence of a bimetallic particle anchored to the silica oxygen atoms through the silver atoms of the cluster. Highresolution electron microscopy of the heattreated material shows a uniform distribution of the bimetallic nanoparticles aligned along the zeolite channels. The catalytic performance of the activated, supported bimetallic particles was tested for hydrogenation of hex-1-ene to hexane. Initial experiments showed a high selectivity (in excess of 99 per cent) and a turnover frequency of at least 6300 mol hexane per mol [Ag,RuIo]per hour. T h e success of these studies led to the incorporation of the copper-ruthenium cluster described preanion, [ {RU,C(CO),,}~CU~CI.']~~, viously (Figure 14) (32), into the mesoporous channels of silica (38). Gentle thermolysis of the anchored clusters gives the bimetallic nanoparticles, characterised by X-ray absorption and FT-IR spectroscopies, and high-resolution scanning transmission electron microscopy. The copper and ruthenium K-edge X-ray absorption spectra show that these catalytically active particles have diameters of approximately 1.5 nm and display a rosette-shaped structure with 12 exposed ruthenium atoms that are connected to a square base composed of relatively concealed copper atoms. In turn, these are anchored by four oxygen bridges to four silicon atoms of the mesopore. The nanoparticles are Fig. 16 The structure of the metal core of dianion: [A~J~&(CO)UC~]*- Platinum Metals Rev., 1998, 42, (4) 155 active catalysts for the hydrogenation of hex1-ene, diphenylacetylene, phenylacetylene, stilbene, cis-cyclooctene and D-limonene, with turnover frequencies of 22400, 17610,70, 150 and 360, respectively, at 373 K and 65 bar of dihydrogen. The catalysts showed no tendency to sinter, aggregate of fragment into their component metals during these experiments. Conclusions In this review it has been shown that synthetic strategies to prepare high nuclearity, bimetallic clusters in good yields have been developed. The metal cores of these clusters have dimensions in excess of 1 nm. By careful control of reaction conditions it is possible to obtain specific target molecules with known ratios of the two metallic components, and the methodology may be extended further to encompass the majority of the late transition elements. In the “condensed” clusters obtained, for the major- ity of the osmiudmercury, rutheniudmercury, osmiumigold and copperiruthenium systems investigated, the mercury, gold or copper atoms form a central domain and the osmium or ruthenium cluster units are fused onto the periphery of these central units. In no case did the two metallic components become dispersed throughout the metal core. Lastly, evidence is beginning to emerge that nanoparticles derived from these and related clusters may prove to be active catalysts when anchored on silica or alumina supports. Acknowledgements My grateful thanks go to Professor the Lord Lewis and Professor Brian F. G. Johnson for their support and encouragement over the years, and for initiating the research described in this review. I am also indebted to the many research workers in the Department of Chemistry, at Cambridge, who have carried out the synthetic and structural work described, and to Johnson Matthey for the generous loan of the heavy transition metal salts. References 1 “Transition “TransitionMetal MetalClusters”, Clusters”, ed. ed. B. B. F. F. G. G.Johnson, Johnson, 8 Wiley, Wiley, New New York, York, 1980; 1980;“Metal “Metal Clusters”, Clusters”, ed. ed. M. M. Moskovits, Moskovits, Wiley, Wiley, New New York, York, 1986; 1986; “The ‘‘The 9 Chemistry Chemistry of of Metal Metal Cluster Cluster Complexes”, Complexes”, eds. eds. D. D. F. F. Shriver, Shriver, H. H. D. D. Kaesz Kaesz and and R. R. D. D. Adams, Adams, VCH VCH Publishers, publishers, Weinheim, Weinheim, !990; 1990; D. D, M. M. P. p, Mingos ~i~~~~ 10 D, J. wales, Introduction cluster and and D. J. Wales, “An Introduction to Cluster Chemistry”, Chemistry”, Prentice-Hall, Prentice-Hall, New New York, York, 1990; 1990; -Clusters and and Colloids. Colloids. From F~~~ Theory Theory to “Clusters Application”, Application”, ed. ed. G. G. Schmid, Schmid, VCH VCH Publishers, Publishers, 1994; “Metal-Metal ,,-,-Weinheim, w ,nhie 1994;C. c,E. E. Housecroft, ~ ~ ~ “Metal-Metal ~ ~ ~ Bonded Carbonyl Carbonyl Dimers Dimers and and Clusters”, Clusters”, Oxford Oxford Bonded University Press, Oxford, Oxford, 1996 1996 University Press, 11 2 G. Dalton G. Schmid, Schmid, J. J.Chem. Chem. SOC., SOC., Dalton Trans., Trans., 1998, 1998, 1077 1077 3 D. D. M. M. P. P. Mingos Mingos and and A. A. S. S. May, May, in in “The “The 12 Chemistry Chemistry of of Metal Metal Cluster Cluster Complexes”, Complexes”, eds. eds. D. D. F. Shriver, H. D. Kaesz and R. D. Adams, VCH F. Shriver, H. D. Kaesz and R. D. Adams, VCH Publishers, Publishers, Weinheim, Weinheim, 1990, 1990, pp. pp. 11-1 11-1 19 19 4 D. D. M. M. Washecheck, Washecheck, E. E. J. J. Wucherer, Wucherer, L. L. F. F. Dahl, Dahl, A. A. Ceriotti, Ceriotti, G. G. Longoni, Longoni, M. M. Manassero, Manassero, M. M. 13 Sansoni Sansoni and and P. P. ChiniJ ChiniJ Ant. Ant. Chem. Chem. SOC., Soc., 1979, 1979, 101,6110 101,6110 5 A. A. Ceriotti, Ceriotti, F. F. Demartin, Demartin, G. G. Longoni, Longoni, M. M. Manassero, M. Marchionna, G. Piva Piva and M. Sansoni, Angew. Chem., Int. Ed. Engl., 1985, 24, 24, 697 14 6 H. Krautscheid, D. D. Fenske, G. Baum and M. Semmelmann, Angew. Chem., Inr. Int. Ed. Ed. Engl., 1993, 32, 1303 15 7 G. Schmid, U. Giebel, W. Huster and A. Schwenk, Inorg. Chim. Acta, 1984,85, 1984, 85, 97 Platinuni Metals Rev., 1998,42, (4) G. Schmid, Chenz. Rev., 1992, 92, 1709 G. Schmid, B. Morun and J.-0.Malm, Angew. Chem., Int. Ed. Engl., 1989, 28, 778 N. M. Vargaftik, I. I. Moiseev, D. I. Kochubey and K. I. Zamaraev, Faraday Discuss. Cheni. SOC., 1991,92, 13; M. N. Vargaftik, V. P. Zagorodnikov, I. P.Stolyarov, I. I. Moiseev, V. A. Likholobov, D. I. Kochubey, A. L. Chuvilin, V. I. Zaikovsky, K. I. Zamaraev ~ ~ andf G. I.t Timofeeva, , j? Chem. Soc., Chenz.C o n z m u n . y 1985, 937 G. Schmid, in “Clusters and Colloids. From Theory to Application”, ed. G. Schmid, VCH Publishers, Weinheim, 1994, p. 178 J. de Jongh, in “Physics and Chemistry of Metal Cluster Compounds. Model Systems for Small Metal Particles. Series on Physics and chemistry of Materials with Low-Dimensional Structures”, ed. J. de Jongh, Reidel, Dordrecht, 1994 M. T. Reetz, W. Winter and B. Tesche, Chenz. Conzmun., 1997, 147; T. Sato, D. Brown and B. F. G. Johnson, Chem. Conznzun., 1997, 1007; B. Dusemund, A. Hoffmann, T. Salzmann, U. Kreibig and G. Schmid, Z. Phys. D , 1991,20,305 L. H. Gade, B. F. G. Johnson, J. Lewis, M. McPartlin, H. R. Powell, P. R. Raithby and W.T. Wong,J. Chem. SOC.,Dalton Trans., 1994, 521 E. L. Muetterties, T. N. Rhodin, E. Band, C. F. Brucker and W. R. Pretzer, Chenz. Rev., 1979,19, 91 156 16 B. F. G. Johnson, J. Lewis, C. E. Housecroft, M. 17 18 19 20 21 22 27 Z. Akhter, Ph. D. Thesis, University of Cambridge, 1995 A. Gallop, M. Martinelli, D. Braga and F. Grepioni,J. Mol. Catal., 1992, 74, 61; B. F. G. Johnson, M. A. Gallop and Y. V. Roberts,J. Mol. Catal., 1994, 86, 51 B. F. G. Johnson and J. Lewis, Adv. Inorg. Chem. Radiochem., 1981,24, 225 E. J. Ditzel, B. F. G. Johnson, J. Lewis, P. R. Raithby and M. J. Taylor, J. Chem. SOC.,Dalton Trans., 1985, 555 J. Lewis, C.-K. Li, M. C . Ramirez de Arellano, P. R. Raithby and W.-T. Wong, J. Chem. Soc., Dalton Trans., 1993, 1359 R. Buntem, J. Lewis, C. A. Morewood, P. R. Raithby, M. C. Ramirez de Arellano and G. P. Shields,J. Chem. SOC.,Dalton Trans., 1998, 1091 D. Braga, P. J. Dyson, F. Grepioni and B. F. G. Johnson, Chem. Re%, 1994,94, 1585; P. R Raithby and G. P. Shields, Polyhedron, 1998, in press I. D. Salter, A d v . Organomet. Chem., 1989, 29, 28 V. Dearing, S. R. Drake, B. F. G. Johnson, J. Lewis, M. McPartlin and H. R. Powell, J. Chem. SOC., Chem. Commun., 1988, 1331 29 Z. Akhter, S. L. Ingham, J. Lewis and P. R. Raithby, Angew. Chem., In?. Ed. Engl., 1996, 35, 992 30 M. A. Beswick, J. Lewis, P. R. Raithby and M. C. Ramirez de Arellano, Angm. Chem., Int. Ed. Engl., 1997,36,291 3 1 M. A. Beswick, J. Lewis, P. R. Raithby and M. C. Ramirez de Arellano, Angm. Chem., Int. Ed. Engl., 1997,36,2227 32 M. A. Beswick, J. Lewis, P. R. Raithby and M. C. Ramirez de Arellano, J. Chem. SOC.,Dalton Trans., 1996,4033 33 S. Kawi and B. C. Gates, in “Clusters and Colloids. From Theory to Applications”, ed. G. Schmid, VCH Publishers, Weinheim, 1994, p. 249 299 23 M. Fajardo, H. D. Holden, B. F. G. Johnson, J. 34 J. H. Sinfelt, “Bimetallic Catalysts: Discoveries, Lewis and P. R. Raithby, J. Chem. SOC.,Chem. Commun., 1984,24 24 L. H. Gade, B. F. G. Johnson, J. Lewis, M. McPartlin and H. R. Powel1,J Chem. Sac., Chem. Commun., 1990, 110 25 P. J. Bailey, B. F. G. Johnson, J. Lewis, M. McPartlin and H. R. Powel1,J. Chem. SOC., Chem. Commun., 1989, 1513; P. J. Bailey, M. J. Duer, B. F. G. Johnson, J. Lewis, G. Conole, M. McPartlin, H. R. Powell and C. E. Anson, J. Organomet. Chem., 1990,383,441 26 E. Charalambous, L. H. Gade, B. F. G. Johnson, T. Kotch, A. J. Lees, J. Lewis and M. McPartlin, Angew. Chem., Znt. Ed. Engl., 1990,29, 1137; L. H. Gade, B. F. G. Johnson,J. Lewis, M. McPartlin, T. Kotch and A. J. Lees,J. Am. Chem. SOC.,1991, Concepts and Applications”, Exxon Monograph, Wiley, New York, 1983 35 M. S. Nashner, D. M. Somerville, P. D. Lane, D. L. Adler, J. R. Shapley and R. G. Nuzzo, J. A m . Chem. SOC.,1996, 118, 12964 36 M. S. Nashner, A. I. Frenkel, D. L. Adler, J. R. Shapley and R. G. Nuzzo, J. A m . Chem. SOC., 1997,119,7760 37 D. S. Shephard, T. Maschmeyer, B. F. G. Johnson, J. M. Thomas, G. Sankar, D. Ozkaya, W. Zhou, R. D. Oldroyd and R. G. Bell, Angew. Chem., Inz. Ed. Engl., 1997, 36, 2242 38 D. S. Shephard, T. Maschmeyer, G. Sankar, J. M. Thomas, D. Ozkaya, B. F. G. Johnson, R. Raja, R D. Oldroyd and R. G. Bell, Chem. Eur. 3,1998, 4, 1214 113,8698 Construction of Miniature Organo-Rhodium Boxes I Many types of molecular cages exist in which ions, atoms or molecules can be trapped. These cages are usually held in suspension and are typically constructed from bifunctional ligands, with square planar or tetrahedral metal centres at the vertices. Until now there have been no cubic shaped organometallic cages. However, if octahedral transition metal building blocks could be constructed, then the assembly of cubicshaped structures should be possible. Now, researchers fi-om the University of Illinois have succeeded in constructing a molecular box from a cubic array of cyano-linked rhodium and cobalt octahedra (K. K. Klausmeyer, T. B. Rauchfuss and S. R. Wilson, Angew. Chem. Znt. Ed., 1998, 37, (12), 1694-1696). Tricyanometalates Et4N[Cp’Rh(CN),] and K[CpCo(CN),] (where Cp‘ = C5Me5,Cp = Platinum Metals Rev., 1998, 42, (4) 1 C5H5)were used to prepare a series of molecular “squares”, by reaction with [Cp’RhCl,] or [(cymene)RuCl,] (cymene = 4-isopropyltoluene). To assemble the box from the “squares” the chloride ligands were removed by AgPF,. The “molecular boxes” of most interest have the structure [(C5R5)8M8(p-CN),z] (M = Rh or Co) and are a subunit of hexacyanometalates, of which Prussian blue is one example. The most interestingbox has alternate rhodium and cobalt atoms at the vertices, linked by CN groups. Each metal atom can adopt its preferred octahedral position. The box has edges 5.1 A long with a volume of 132 A’, giving enough space inside to encapsulate a caesium atom. The box is also soluble so it could therefore be used for trapping molecules in solution. - 157